1. Field of the Invention
The present invention relates to a method for encoding a depth image using color image information and a method for reconstructing the depth image using the color image information, and more particularly, to a method for selecting a sampling value from the depth image and encoding the sampling value using the color image information and mesh-based method for reconstructing a depth image using the color image information.
2. Discussion of Related Art
Recently, depth image-based rendering (DIBR) has been spotlighted as a method of representing a 3-dimensional video. DIBR is a method for creating scenes at different viewpoints using reference images having information such as a depth, a difference angle, etc. for each pixel related. Such DIBR not only easily performs rendering on shapes which are difficult and complex to represent using a 3-dimensional model, but also enables an application of a signal processing method such as general image filtering. In addition, such DIBR has an advantage of enabling production of a high-quality 3-dimensional video.
Such DIBR uses a depth image (a depth map) and a texture image (a color image) obtained through a depth camera and a multi-view camera to realize the foregoing. Particularly, the depth image is used for representing the 3-dimensional model more realistically (that is, is used for producing the 3-dimensional video having a greater 3-dimensional effect).
A depth image may be defined as an image which represents a distance between an object located in a 3-dimensional space and a camera photographing the object in black-and-white units. Such a depth image is often used in 3-dimensional restoration techniques or 3-dimensional warping techniques through depth information and camera parameters.
In addition, depth images are applied in various fields such as free viewpoint television and 3-dimensional television. Free viewpoint television is television which allows a user to watch an image at a convenient viewpoint according to the user's selection, rather than watching the image at only one viewpoint which is previously set. This feature of free viewpoint television allows creation of images at random viewpoints in consideration of multiview images photographed by several cameras and multiview depth images corresponding to the multiview images. Meanwhile, 3-dimensional television has a feature of implementing a realistic image by adding a depth image to an existing 2-dimensional television, and active research and development of 3-dimensional television has been performed recently.
Also, an existing method of encoding and reconstructing a mesh-based depth image includes selecting a part in which a high-frequency component is large within color information or a depth image as a position of a standard value. However, since encoding techniques mainly include eliminating or decreasing high-frequency components, encoding and reconstructing of such sampling values tend not to be efficient.
To solve the foregoing problems of the related art, the present invention provides a method for segmenting the depth image into block units and selecting and encoding a sampling value in each block using the color image information, and a method for reconstructing the mesh-based depth image using the color image information.
An object of the present invention is not limited to the foregoing object and the above and other objects, features and advantages of the present invention may become more apparent to those of ordinary skill in the art from the following detailed description.